Groundwater age dating using single and time-series data of environmental tritium in the Moeda Synclyne, Quadrilátero Ferrífero, Minas Gerais, Brazil

Highlights

• Tritium time series provide a mean age.

• The best model was the EPM model.

• Short residence time was found for Cauê aquifer.

Abstract

The use of tritium in the identification of the “age” of groundwater is being used since the discovery of its potential for this purpose in 1954. The present work sought to apply and discuss the use of the different types of lumped parameter models (LPM) in the Cauê and Gandarela aquifers, in order to obtain the groundwater mean residence time based on tritium data. Water samples from several discharges related to the Cauê and Gandarela aquifers were collected during the month of September of 2017. The reconstruction of the historical series of tritium contents in the precipitation at the GNIP station CDTN-Belo Horizonte was carried out based on data from the Kaitoke (New Zealand) and Adelaide (Australia) stations in addition to the Brazilian stations Brasília, Rio de Janeiro and Salvador. The sample results and the data of the reconstructed historical series of the tritium concentration in the CDTN station during the period between 1953 and 2017, served as the basis to use the lumped parameter models with the aid of the software TracerLPM, provided by the United States Geological Survey. From the overlap of some sample points with tritium samples collected in different years by different authors, an adjustment of some models of residence time distribution was possible. The models used were exponential (EM), exponential-piston (EPM), and dispersive (DM). The model that best fit to data was the EPM model. The mean transit time for the EPM model in the analyzed points varies from 63 to 95 years, with a median value of 73 years. Differences between values of the EPM ratio parameter were verified. However, all values indicate an exponential flow component greater than the piston flow component. The EPM model suggests a long retention time of the tracer in the aquifer system. It can be used as an approximation of the retention time for environmental contamination. The exponential model (EM) and piston flow model (PFM) were used together. Based on the data of a single tritium sample collected in time, and the average value of residence time provided by each model, a first approximation of the mean residence time value can be obtained. The results suggest that this is a better approximation than assuming a pure exponential model or piston flow model to describe the data of a single tritium sample in time as were made by previous local works. In general, a short residence time was estimated for the Cauê aquifer as a whole with some discharges with long residence time that drains regional flows from the deepest part of the Cauê aquifer, reflecting the complex structural geology fabric.

Introduction

Groundwater systems frequently have different flow paths that follow the direction of a system outflow (spring or well). For this reason, a single water sample does not have a unique age. Instead of it, the water sample has a distribution of ages, which can be described by flow models that represent idealistic aquifer conditions (Stewart, 2012; Suckow, 2014). The groundwater age (age and residence or transit time are used here as synonymous) using tritium as a tracer has been used since the discovery of its potential by Kaufman and Libby (1954). The work of Nir (1964) proposed a mathematical approach to interpret the radioactive isotope data using the piston flow model (PFM), exponential model (EM), and the dispersive model (DM). These earlier models were continuously improved in works such as Maloszewski and Zuber (1996), Amin and Campana (1996) and Zuber and Maloszweski (2001). They present different kinds of models to simulate specifics hydrogeological conditions, such as linear model (LM), exponential-piston model (EPM), linear-piston model (L-PM), among others. These types of models are known as Lumped Parameters Models (LPM) and involve using equations that consider the aquifer system as a whole to describe the residence time distribution of the aquifer. Using the convolution equation, it is possible to estimate the concentration and time in the system output by knowing the time and concentration of any input tracer.

Mean residence times and residence time distribution functions are remarkable results supplied by applying Lumped Parameters Models. These results can be used to describe transport, mixing, and storage of groundwater and surface water systems and therefore about the retention and release of contaminants (Stewart et al., 2017). The information obtained from isotopic data is not only an important input for local water management regulatory agencies, but it also supplies essential information to understand the impact of climate change on water resources.

References about the use of Lumped Parameters Models and isotopic data are vast. Maloszewski et al. (2002) used tritium and oxygen-18 data modeling to identify the flow system of the karstic-fissural-porous aquifer called Schneealpe (Austria), composed of limestone and dolomites from the Triassic and which is used as a drinking water supply of the city of Vienna. A dispersive model was applied to interpret the tritium data of the porous fissure system during periods of drought (baseflow), and a piston flow model was used to treat the oxygen-18 data that oscillate seasonally due to the presence of karst ducts that connect the aquifer to the surface. The work of Stewart and Morgenstern (2016) presents the importance of tritium based transit times in hydrological systems to determinate the parcel of the hidden streamflow in river catchments. Furthermore, the authors encourage the catchment community to continue to sample for tritium concentration determination in order to obtain good time series for modeling input. A demonstration about the application of tritium to estimate groundwater transit times and storage volume in a Japan watershed can be seen in Gusyev et al. (2016). The authors use rivers tritium samples and an EPM model with a piston flow component of 30% to obtain the groundwater mean transit time and baseflow data to calculate the groundwater storage volume and the mean saturated thickness.

The study area of this work is marked by water conflicts and deficient hydrogeological studies, many of which are concentrated in mining areas and are only related to dewatering studies. The area is located at the west limb of a regional syncline (named as Moeda Syncline), in the west portion of the geological province known as Quadrilátero Ferrífero, Minas Gerais State. The Cauê aquifer is an important aquifer for water supply and is recognized by the massive amount of banded iron formations (named as itabirite). The complex structural geology involving the rocks from this aquifer, as described in Chemale et al. (1994), and usually high layer dip angles result in a complex groundwater system. Previous works on hydrogeology discipline poorly explored the isotopic information, neglecting important information on tritium modeling. The objective of this work is to use the lumped parameter model approach using tritium data to obtain the mean transit time and residence time distribution model for the Cauê aquifer discharges. Also, this work develop a discussion about the hydrogeological conceptual model and modeling uncertainties.